• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Core A2 version 2.10

[H]ugh_Freak

[H]ugh_Freak

[H]ard|Gawd
2FA
Joined
Apr 30, 2003
Messages
1,133
Anyone else pick up these today?

So far my VMs on a 3.2GHz Q6600 have gone from 8-9 minutes a frame to 13-14 minutes. On a 4 core Hyper-V instance it went from 5-6 minutes to 7-8 minutes. If you update the core during a WU it'll kill the WU and download a new one or 5. Or at least one of my machines did. Downloaded 5 before it found one that would run with the new core. So far the only good thing I've seen is that the core actually auto-terms when it comes up with a bad WU rather then hanging like 2.07 or 2.08. Oh and if I tried to replace 2.10 with 2.07 or 2.08 it would say the core is outdated and download 2.10 again.
 
yeah .. performance of the new core does suck .. at least some of the bad units process .. planning to give this a few days .. if the situation doesn't improve, it might be time to move my last box over to WCG .. it's not worth running LinSMP on dual core boxes anymore .. only getting 1100PPD at the moment .. none of these problems happen with BOINC ;)
 
I've dropped from just under 3k to 2300 ppd on my main rig. Running a VM with notfreds. Still worth me running the client. I've gotten to the point that ppd are nice, but the science is what matters.
 
I've dropped to around 2000-2100PPD on my notfred's VM from about 2700-3000 depending on what I'm doing on my PC. My Fedora VM is still running the 2.08 core and I plan on keeping it that way for the time being. Hopefully they'll bring out another version of the core in the near future that brings the performance back up to where it used to be. Right now it's not doing much better than the a1 core.
 
I believe 2.10 is ment to correct the 1 core WU bug going around ...

I have gotten quite a few of those the last few days before I shut down my farm

I am sure they will address the other issue as fast as they can

I would also guess most of the Stanford's time is going toward SMP2 testing
 
Evil- do you have any idea how far away they are from releasing an SMP2 client?
 
one more here wondering about SMP2, that may help us get more people back into folding
(ooo, come fold for us, get all these flashy points!!)

ok, maybe not, but im still curious
 
Yea I would definately like some deets on SMP2


/me whispers to Evil......we have private messages;)
 
its gotta be soon, I mean with all the power thats out there now, think of what a new SMP could yield for stanford, theyd go crazy with data! That and the new grfx cards coming in a few months, I hope stanford capitalizes on this stuff
 
Sometimes I wonder though if stanford can really do anything with all this data?
Or if they even have enough WU's to send out.

I mean I look at this new core....same WU's but lower PPD? I mean do you think they did this to slow us down intentionally?

Maybe im just being skeptical....:confused:
 
Looks like my Fedora VM automatically downloaded the new core, so now both VMs are down to about 2000PPD each :(.
 
[H]ugh_Freak;1034573454 said:
Anyone else pick up these today?

So far my VMs on a 3.2GHz Q6600 have gone from 8-9 minutes a frame to 13-14 minutes. On a 4 core Hyper-V instance it went from 5-6 minutes to 7-8 minutes. If you update the core during a WU it'll kill the WU and download a new one or 5. Or at least one of my machines did. Downloaded 5 before it found one that would run with the new core. So far the only good thing I've seen is that the core actually auto-terms when it comes up with a bad WU rather then hanging like 2.07 or 2.08. Oh and if I tried to replace 2.10 with 2.07 or 2.08 it would say the core is outdated and download 2.10 again.
Click to expand...
Everything you mentioned including the deletion of a WU in mid-progress when DL the new core, I have confirmed. The main difference I'm experiencing from most of you is the amount of decrease in performance. It ranges from none to a maximum of 14% decrease on the worst affected VMs.

The P2677 is seeing the greatest decrease whereas other WUs are not decreasing their performance that much. The good thing about this core is that it seems to have ended the dreaded slowdown I was experiencing with recent corrupt WUs that prompted constant monitoring. I've had so many corrupt WUs lately that any little slowdown will not diminish my production much since I won't need to delete and DL new WUs.

I'm certain Stanford updated the core to rectify this situation. I'm hoping that in a week or two this core will be replaced with something a little faster but for now, I'm glad I won't need to constantly monitor my clients. That was getting a little out of hand lately.
 
Last edited:
APOLLO said:
Everything you mentioned including the deletion of a WU in mid-progress when DL the new core, I have confirmed. The main difference I'm experiencing from most of you is the amount of decrease in performance. It ranges from none to a maximum of 14% decrease on the worst affected VMs.

The P2677 is seeing the greatest decrease whereas other WUs are not decreasing their performance that much. The good thing about this core is that it seems to have ended the dreaded slowdown I was experiencing with recent corrupt WUs that prompted constant monitoring. I've had so many corrupt WUs lately that any little slowdown will not diminish my production much since I won't need to delete and DL new WUs.

I'm certain Stanford updated the core to rectify this situation. I'm hoping that in a week or two this core will be replaced with something a little faster but for now, I'm glad I won't need to constantly monitor my clients. That was getting a little out of hand lately.
Click to expand...

I'm getting slightly different PPD decreases depending on the WU. I'm also pretty sure core 2.10 was released to fix all the buggy WUs. Yeah, the constant monitoring the last couple weeks was a big pain. I'm hopeful the production will get better soon as well. I'm willing to sacrifice some performance for better stability but I've gotten slowdowns near 1/3rd on some WUs.
 
nitrobass24 said:
Yea I would definately like some deets on SMP2
/me whispers to Evil......we have private messages;)
Click to expand...

hahaha .... i am sorry i don't have any more info then what is aleady out there ...

The only two places Stanford has even brought it up is Pande's Twitter and the F@H News Page ...
 
I woke up this morning to a A2 unit hung for two hours at 0% complete. The previous A2 unit completed successfully with the new core version.

After almost a hour and 10 failed attempts to get a good unit, my notfred box got switched to WCG. I plan on running 100% WCG work until these issues get worked out.

Stanford really needs to address this issue or give a resonable explanation why it's happening ... the 2.10 core upgrade obviously doesn't fix the problem.
 
AgrFan said:
Stanford really needs to address this issue or give a resonable explanation why it's happening ... the 2.10 core upgrade obviously doesn't fix the problem.
Click to expand...
Interesting. I'm going to have to check my VMs a little more often just in case this happens to me. I thought we would have a respite from monitoring but it looks like a bit more attentiveness might be required.

I honestly thing the v2.10 core is a temporary solution until Stanford releases a newer, faster and stabler core. It's unfortunate this happened but the LinSMP client has been very stable for a number of years. I had a problem last year around this time when I was one of the few people experiencing hangs right after a WU was completed and uploaded results. The client just stalled and wouldn't DL a new WU. Some people may recall my complaints last year. Stanford released a beta client to alleviate the problem. They do get things fixed but it can can take some time.
 
Any chance the new 2.10 core has higher memory requirements? I'm running notfred from a USB stick on a Dell E521 with a AMD 3800+ and 1 GB of memory. FAHLog.txt shows 937mb being reported. That still doesn't explain why I was getting units with download sizes in the 1-2mb range. My understanding is that the new core will dump units until a good one is downloaded. That wasn't happening consistently this morning. Some units would dump and others would immediately hang the box. I did complete two A2 units with the new core previously. Could it be possible I'm out of horsepower for the SMP client?
 
Last edited:
AgrFan said:
Any chance the new 2.10 core has higher memory requirements? I'm running notfred on a Dell E521 with a AMD 3800+ and 1 GB of memory. I did complete two units with the new core before today. That wouldn't explain why I was getting units with download sizes in the 1-2mb range though. My understanding is thatthe new core will dump units until a good units is downloaded. That wasn't happening consistently this morning.
Click to expand...
I honestly don't know. Perhaps Evil has more information?
 
grr cant wait for the CPU to make a comeback... killing me having all this GPU power and the CPU cant even compare
 
Vaulter98c said:
grr cant wait for the CPU to make a comeback... killing me having all this GPU power and the CPU cant even compare
Click to expand...
You and me both. With all the CPUs I have here (four 8-cores) my production is underrepresented, but hey, I go with the flow and if Stanford wants to emphasize GPU, I'm cool with that. I won't be building many mega-core systems any time soon, though, if that's the general trend for the future of folding. ;)
 
My PpD is definitely down. In addition, I am getting EUE's on one box when I previously wasn't having any issues :( :(

I am still having to babysit, just not as frequently. The box that I am getting EUE's on, may be moved to WCG. I have two rigs moved over to WCG form F@H SMP and frankly that may not change.
 
AgrFan said:
Interesting read ...

http://en.fah-addict.net/news/news-0-20+a-focus-on-the-smp-a2-2-10-2-11-core.php
Click to expand...
Cool, a v2.11 core is already being tested on Macs. It might soon become available then. I also noticed the increased CPU usage of the v2.10 core. My single-core clients are now hardly producing anything at all. The slower performance of this core and its greater CPU utilization indicates how much less efficient it is compared to previous cores. No doubt the efficiency issue will be addressed in the next core.
 
You must log in or register to reply here.
Back
Top